Immunology Today, vol. 5, No. 8, 1984
222 sequences of the latter two were not available at the time our review was written. The point of variance between ourselves and Williams is that we coneluded that the evidence for considering Thy- 1 an immunoglobulin homolog was not compelling. Williams argues that other parameters such as shape, patterned sequence and presence of characteristic cysteine residues should also be considered. We agree, and pointed out that the generally accepted belief that Ig V and C regions share a distant common ancestry is based on such criteria, inasmuch as sequence comparisons are seldom statistically significant. He incorrectly claims that we have not considered domain homology; in fact, all of our sequence comparisons were made using domain-sized regions (see Table III of Ref. 1). In this table, we show using R E L A T E that the Ig-like domains of M H C products, HLA-B7, H L A - D r a and HLA-Drfl show a strong statistical relatedness to Ig constant regions that is actually more 'significant' than the generally accepted comparison with
F u r t h e r c o m m e n t s o n the molecular members of extended immunoglobulin family
SIR, We are grateful for this opportunity to reply to A. F. Williams because it allows us to clarify our major points regarding immunoglobulin (Ig)-related molecules. Upon reading Williams' comments on 'molecules in the immunoglobulin superfamily', one might get the impression that there is a great discrepancy regarding what we said 1in our review on
Probing the Boundaries of the Extended Immunoglobulin Family and Williams' viewpoint. We are in essential agreement that the proper members of the immunoglobulin superfamily at this time include immunoglobulins and M H C antigens, and we would also include the products of the putative T-cell receptor genes 2-5 and the poly-Ig receptor 6 although the
fl2-microglohulin. Thy 1. i, al-acid glycoprotein and C-reactive protein did not show acceptable levels of significance with V or C region domains. One of the major conclusions we drew is that proteins such as products of the major histocompatibility complex and immunoglobulins which on a 'bulk comparison basis' are marginally related if at all, show strong segmental homology in certain domain sized regions. This was the basis for the 'jumping domain' concept we proposed. Furthermore, we provide evidence based upon serology and comparison of short stretches of sequence showing that molecules which are not immunoglo'bulins can nevertheless have combining site regions serologieally related to immunoglobulin idiotopes which must, therefore, have a common steric shape although the sequence identities are limited to short stretches. In this case, it is possible to argue for some sort of minigene insertion model or for convergence based upon sterie requirements for forming a proper combining site. The point o four exercise
TABLE I. Segmental comparisons of framework regions of MOLT-3 putative T cell receptor gene with sequences of cross-reactive immunoglobulin V H. The amino acid sequence corresponding to the gene sequence of the YT35 T cell receptor gene was taken from reference 2. The variable region sequences for the immunoglobulin chains was taken from reference 12. Framewo~ 1 YT35 McE TEPC - 15 MOPC-104E MOPC-315
YT35 McE TEPC- 15 MOPC-104E MOPC-315
1QI 1EV 1EV 1DV
T K Q Q
L K E S G p T L V K r T E T LmLIT L V E S G G G L V Q p~G S L R~ S L~Q S G p~L V K p~A S~KM S L E S G P G L V K P~[]S~S[]T~S
Framework 2 53'[-~Y[-R-~ T MM R G[LE] L~]
36]W[IlRIO[R P G K 361wlvIR[(~lp p G U
T F
V
T~Y
~ T F S T F []I T
I
AlL EIW~A NIL E[W I A
3 6 1 W [ V ~ J ( ~ I S H G K S [ L EIW I G
Framework 3 YT35 McE TEPC - 15 MOPC- 104E MOPC-315
6 6 [ R L~L T 66~AF I 66 Z 66[~V S
G T D T S R N QV V V S R D T S SQ~I L Y L T V D K S
YT35 McE TEPC- 15 MOPC-104E MOPC-315
Q L N S L T S
i T R D T S E N Q F F[~K-]L Framework 4 ~J)
126 F GS 103 W GQ 103 W GA 103 W[GA 103
T A g g
N M D P~JV D S ~ T I Y ~ C ] A [ N A L R A E[D T ~ A ~ _ ~ Y [ Y ] C I A ] R
GTR~TVVE GTL~TVSS GTTV[TVSS GTT~TVSS
WlGQ G T T ~ T V S
D S V T
E D[p__~AV~YIYICIAIR T A[AJTY[_YJYC[~_AJG
223
Immunology Today, vol. 5, No. 8, 1984
TABLE II. Sequence comparison of routine Thy-l.2 with rabbit and chicken skeletal muscle actin and rabbit vimentin showing regions of homology reported by Dales et al. 11*
Thy-l.2 Actin
1 5 10 Z K V T S L ~ _ C ~ N ID E D E T
86 Thy-1.2
Actin Vimentin
L 00
N E L R E
90
S
95
15 Q~L C D G 100
105
A K
110
n
P L .... [ ~ ] F
64 70 N~P Y I K y ~ 58A S K R G
S [~]L[N-~L
K RLRJT L
*Dales et al. n followed computer analyses using programmes MATCH, SEARCH and ALIGN and selected segments visually that showed maximum homology. was not a slavish rehashing of computer analyses but an attempt to (!) define relationships and to (2) infer novel genetic mechanisms, e.g. 'jumping domains' and 'minigene insertion', which might occur in the evolution of the immunoglobulin-related family of recognition molecules. The T-ceil receptors turn out to be an interesting case because these molecules, based upon serology using antibodies to defined regions ofV H structure 7-9and by comparison of gene sequence 2'3 show segmental homology with immunoglobulins, and the identification of discrete regions corresponding to J, D and the third framework. Moreover, they can be readily aligned with variable regions to show the existence of other frameworks and complimentarity determining regions. When, in fact, regions of amino acid sequence corresponding to V T frameworks (Frl, Frz, Fr 3 and JT) are aligned with classical Vx regions known to be serologically crossreactive in some segments, e.g. JH (Ref. 7), identities in the range of 40-50 % are found (Table I). This observation indicates that the putative T-cell receptor genes fit within the Ig-family, as opposed to marginally related molecules. The arrangement of gene segments, i.e. V, D,J, C, is consistent with that of heavy chains 10 although the VTC T system represents a new translocon. Hedrick et al. s 'conclude that the locus described (here) represents a type of immunoglobulin gene specifically rearranged and expressed by at least some subsets of T lymphocytes'. It is interesting that the V T sequence of the YT35 gene product as shown here suggests the existence of a minigene for CDR1 because the entire region corresponding to this segment has been deleted from the V T structure. RegardingThy 1 as an immunoglobulin homolog, Williams claims that 'the similarity between Thy-1 and Vdomains of N E W M heavy and light
chains was most unlikely to be due to chance'. This depends on what you take as the significant cut-off, We might be accused of excess rigor because we took the strict statistical cut-off at the probability of
222 sequences of the latter two were not available at the time our review was written. The point of variance between ourselves and Williams is that we coneluded that the evidence for considering Thy- 1 an immunoglobulin homolog was not compelling. Williams argues that other parameters such as shape, patterned sequence and presence of characteristic cysteine residues should also be considered. We agree, and pointed out that the generally accepted belief that Ig V and C regions share a distant common ancestry is based on such criteria, inasmuch as sequence comparisons are seldom statistically significant. He incorrectly claims that we have not considered domain homology; in fact, all of our sequence comparisons were made using domain-sized regions (see Table III of Ref. 1). In this table, we show using R E L A T E that the Ig-like domains of M H C products, HLA-B7, H L A - D r a and HLA-Drfl show a strong statistical relatedness to Ig constant regions that is actually more 'significant' than the generally accepted comparison with
F u r t h e r c o m m e n t s o n the molecular members of extended immunoglobulin family
SIR, We are grateful for this opportunity to reply to A. F. Williams because it allows us to clarify our major points regarding immunoglobulin (Ig)-related molecules. Upon reading Williams' comments on 'molecules in the immunoglobulin superfamily', one might get the impression that there is a great discrepancy regarding what we said 1in our review on
Probing the Boundaries of the Extended Immunoglobulin Family and Williams' viewpoint. We are in essential agreement that the proper members of the immunoglobulin superfamily at this time include immunoglobulins and M H C antigens, and we would also include the products of the putative T-cell receptor genes 2-5 and the poly-Ig receptor 6 although the
fl2-microglohulin. Thy 1. i, al-acid glycoprotein and C-reactive protein did not show acceptable levels of significance with V or C region domains. One of the major conclusions we drew is that proteins such as products of the major histocompatibility complex and immunoglobulins which on a 'bulk comparison basis' are marginally related if at all, show strong segmental homology in certain domain sized regions. This was the basis for the 'jumping domain' concept we proposed. Furthermore, we provide evidence based upon serology and comparison of short stretches of sequence showing that molecules which are not immunoglo'bulins can nevertheless have combining site regions serologieally related to immunoglobulin idiotopes which must, therefore, have a common steric shape although the sequence identities are limited to short stretches. In this case, it is possible to argue for some sort of minigene insertion model or for convergence based upon sterie requirements for forming a proper combining site. The point o four exercise
TABLE I. Segmental comparisons of framework regions of MOLT-3 putative T cell receptor gene with sequences of cross-reactive immunoglobulin V H. The amino acid sequence corresponding to the gene sequence of the YT35 T cell receptor gene was taken from reference 2. The variable region sequences for the immunoglobulin chains was taken from reference 12. Framewo~ 1 YT35 McE TEPC - 15 MOPC-104E MOPC-315
YT35 McE TEPC- 15 MOPC-104E MOPC-315
1QI 1EV 1EV 1DV
T K Q Q
L K E S G p T L V K r T E T LmLIT L V E S G G G L V Q p~G S L R~ S L~Q S G p~L V K p~A S~KM S L E S G P G L V K P~[]S~S[]T~S
Framework 2 53'[-~Y[-R-~ T MM R G[LE] L~]
36]W[IlRIO[R P G K 361wlvIR[(~lp p G U
T F
V
T~Y
~ T F S T F []I T
I
AlL EIW~A NIL E[W I A
3 6 1 W [ V ~ J ( ~ I S H G K S [ L EIW I G
Framework 3 YT35 McE TEPC - 15 MOPC- 104E MOPC-315
6 6 [ R L~L T 66~AF I 66 Z 66[~V S
G T D T S R N QV V V S R D T S SQ~I L Y L T V D K S
YT35 McE TEPC- 15 MOPC-104E MOPC-315
Q L N S L T S
i T R D T S E N Q F F[~K-]L Framework 4 ~J)
126 F GS 103 W GQ 103 W GA 103 W[GA 103
T A g g
N M D P~JV D S ~ T I Y ~ C ] A [ N A L R A E[D T ~ A ~ _ ~ Y [ Y ] C I A ] R
GTR~TVVE GTL~TVSS GTTV[TVSS GTT~TVSS
WlGQ G T T ~ T V S
D S V T
E D[p__~AV~YIYICIAIR T A[AJTY[_YJYC[~_AJG
223
Immunology Today, vol. 5, No. 8, 1984
TABLE II. Sequence comparison of routine Thy-l.2 with rabbit and chicken skeletal muscle actin and rabbit vimentin showing regions of homology reported by Dales et al. 11*
Thy-l.2 Actin
1 5 10 Z K V T S L ~ _ C ~ N ID E D E T
86 Thy-1.2
Actin Vimentin
L 00
N E L R E
90
S
95
15 Q~L C D G 100
105
A K
110
n
P L .... [ ~ ] F
64 70 N~P Y I K y ~ 58A S K R G
S [~]L[N-~L
K RLRJT L
*Dales et al. n followed computer analyses using programmes MATCH, SEARCH and ALIGN and selected segments visually that showed maximum homology. was not a slavish rehashing of computer analyses but an attempt to (!) define relationships and to (2) infer novel genetic mechanisms, e.g. 'jumping domains' and 'minigene insertion', which might occur in the evolution of the immunoglobulin-related family of recognition molecules. The T-ceil receptors turn out to be an interesting case because these molecules, based upon serology using antibodies to defined regions ofV H structure 7-9and by comparison of gene sequence 2'3 show segmental homology with immunoglobulins, and the identification of discrete regions corresponding to J, D and the third framework. Moreover, they can be readily aligned with variable regions to show the existence of other frameworks and complimentarity determining regions. When, in fact, regions of amino acid sequence corresponding to V T frameworks (Frl, Frz, Fr 3 and JT) are aligned with classical Vx regions known to be serologically crossreactive in some segments, e.g. JH (Ref. 7), identities in the range of 40-50 % are found (Table I). This observation indicates that the putative T-cell receptor genes fit within the Ig-family, as opposed to marginally related molecules. The arrangement of gene segments, i.e. V, D,J, C, is consistent with that of heavy chains 10 although the VTC T system represents a new translocon. Hedrick et al. s 'conclude that the locus described (here) represents a type of immunoglobulin gene specifically rearranged and expressed by at least some subsets of T lymphocytes'. It is interesting that the V T sequence of the YT35 gene product as shown here suggests the existence of a minigene for CDR1 because the entire region corresponding to this segment has been deleted from the V T structure. RegardingThy 1 as an immunoglobulin homolog, Williams claims that 'the similarity between Thy-1 and Vdomains of N E W M heavy and light
chains was most unlikely to be due to chance'. This depends on what you take as the significant cut-off, We might be accused of excess rigor because we took the strict statistical cut-off at the probability of
